Self-hosted · $0/video possible · 21 agents

Type a topic. Get a video.

VigenFlow is an AI pipeline that turns any topic into a publish-ready YouTube video — research, script, voice, video per scene, consistent characters, subtitles, SEO, thumbnail and shorts. Fully automated. Runs on your machine.

21 production agents EN + Vietnamese YouTube · TikTok · Reels
vigenflow

How it works

Three steps to a finished video

No UI needed. No cloud subscription. Your machine, your keys, your content.

✍️Step 01

Type your topic

One command. VigenFlow researches angles, pain points, and trending hooks for your topic automatically.

python main.py "She hid a second family for 7 years" \
  --channel pipeline/toldback/profile.yaml
🤖Step 02

Pipeline runs end-to-end

10 agents run in sequence — each reviewed by a dedicated AI reviewer with retry logic before moving to the next step.

Research → Script → QA → Images → Voice
→ Subtitles → Video → Effects → SEO
→ Thumbnail → Keyword Overlay → Shorts
🚀Step 03

Get publish-ready output

Final video, 3 short clips, thumbnail, SEO metadata, chapters — all scored and ready to upload.

✓ Score: 8.9/10 — Ready to upload
✓ video:     final_video.mp4  (44.2 MB)
✓ shorts:    hook / climax / moral
✓ thumbnail: thumbnail.png
✓ seo:       seo.json

Features

One command. Publish-ready video.

21 agents, each independently reviewable, retryable, and replaceable. Review and regenerate any scene before final export.

🔁Unique

AI Review Loop

Every step has a dedicated reviewer agent. Script, images, audio, thumbnail — all scored and auto-retried up to 3×. No other tool does this.

🎬Unique

Video Per Scene

Each script scene gets its own image sequence, voice clip, subtitle block, and timing — assembled into a full cinematic video.

👤Unique

Character Consistency

Seed a character once per channel. Same face, same visual style, across every scene of every video — automatically.

🔄

Review & Regenerate

Not happy with a scene? Review output, regenerate a single image or scene, and re-assemble. Full control at every step.

🔗

Multi-LLM Chain

Claude → Gemini → GPT-4o → Grok. Auto-fallback per step with no provider lock-in. Free tier first.

🎙️

Voice Chain

ElevenLabs → OpenAI TTS → Kokoro (local, free) → Edge → gTTS. Always has a voice — even offline.

🖼️

Image Chain

Imagen 3 → DALL-E 3 → Pollinations → Pixabay → Pexels. 8-provider cascade. Always has an image.

📱

Auto Shorts

3 vertical clips (hook / climax / moral) with karaoke-style yellow word highlight. Sized for TikTok, Reels, YouTube Shorts.

📊Unique

Score Card

Every run scored on hook, CTA, arc, atmosphere, tension. Monetization risk score included. Engagement prediction per video.

🌏

Multilingual

Full Vietnamese pipeline: FPT TTS, NotoSans font fallback, VI-aware prompts. Add any language via profile.

Channel Profiles

YAML config per channel — visual style, voice, review thresholds, SEO keywords. One base profile, unlimited channels.

🔒

Self-hosted

Runs entirely on your machine. Your API keys, your data, your content. $0/video with free-tier providers.

$3.4B
AI video market by 2033
from $788M in 2025
12h+
Saved per video
vs. manual editing
$80K–300K
Monthly revenue
top faceless channels
38%
New channels faceless in 2025
up from 12% in 2022

Why VigenFlow

Not another clip tool.

Most tools clip, caption, or repurpose existing footage. VigenFlow creates from scratch — a full production pipeline in one command.

Review loop nobody else has

Every agent has a dedicated reviewer. Script scored on hook, arc, and CTA. Images checked for watermarks, duplicates, and visual consistency. Failed steps auto-retry up to 3×.

Review & regenerate any scene

Not happy with a scene's image or voice? Review the output, regenerate just that scene, and re-assemble. Full manual control — no need to restart the whole pipeline.

Character consistency across videos

Seed a character once per channel profile. Same face, same visual style, every scene of every video — works with Imagen 3 and Pollinations FLUX.

Your keys. Your machine. Zero subscription.

$0/video is real: Gemini free tier for LLM, Kokoro (local) for voice, Pollinations (free) for images. Paid providers optional — and you bring your own keys.

Compare

How we stack up

No other tool combines full pipeline, AI review loop, scene-level video, and $0/video option in one self-hosted package.

ToolPriceFull pipelineAI review loopMulti-LLMScene-level videoCharacter consistencyShorts genSelf-hostedVietnameseFree tier
VigenFlow
$0*
InVideo AI
$28/mo
Faceless.video
$19/mo
AutoShorts
$19/mo
Pictory
$19/mo
HeyGen
$24/mo
Synthesia
$29/mo

* $0/video with Gemini free tier + Kokoro (local TTS) + Pollinations (free images). Paid providers optional.

Use Cases

Any channel. Any niche.

Drama, finance, wellness, animation, documentary — VigenFlow adapts to your channel through a YAML profile, not a template picker.

Drama / True Crime

Faceless storytelling channels

Input a story angle → 15-scene cinematic video, consistent characters, male voice, karaoke subs. 3 viral shorts auto-clipped and sized for TikTok.

  • 15 scenes / video
  • ~4 min production
  • 3 shorts auto-clipped
Any Topic

Finance, wellness, history, animation...

VigenFlow works for any niche — one consistent on-screen character explaining finance, an animated narrator, a documentary host. Profile YAML controls the whole identity.

  • Consistent character per channel
  • Channel profile YAML
  • Any language
Agencies & Power Users

Multi-channel production at scale

Run parallel pipelines per channel. Each with its own voice, visual style, LLM chain, and review thresholds. Score card tracks quality across every run.

  • Parallel channel runs
  • Per-channel profiles
  • Full score card

Early Access

Request a Demo

We're onboarding select creators and agencies. Tell us about your channel and we'll reach out to set up a live walkthrough.

No spam. No sales pitch. Just a real walkthrough of the tool.